Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.5/dft/simd/common/n2fv_4.c @ 84:08ae793730bd

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Add null config files
author Chris Cannam
date Mon, 02 Mar 2020 14:03:47 +0000
parents 2cd0e3b3e1fd
children
rev   line source
Chris@42 1 /*
Chris@42 2 * Copyright (c) 2003, 2007-14 Matteo Frigo
Chris@42 3 * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
Chris@42 4 *
Chris@42 5 * This program is free software; you can redistribute it and/or modify
Chris@42 6 * it under the terms of the GNU General Public License as published by
Chris@42 7 * the Free Software Foundation; either version 2 of the License, or
Chris@42 8 * (at your option) any later version.
Chris@42 9 *
Chris@42 10 * This program is distributed in the hope that it will be useful,
Chris@42 11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
Chris@42 12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
Chris@42 13 * GNU General Public License for more details.
Chris@42 14 *
Chris@42 15 * You should have received a copy of the GNU General Public License
Chris@42 16 * along with this program; if not, write to the Free Software
Chris@42 17 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
Chris@42 18 *
Chris@42 19 */
Chris@42 20
Chris@42 21 /* This file was automatically generated --- DO NOT EDIT */
Chris@42 22 /* Generated on Sat Jul 30 16:40:07 EDT 2016 */
Chris@42 23
Chris@42 24 #include "codelet-dft.h"
Chris@42 25
Chris@42 26 #ifdef HAVE_FMA
Chris@42 27
Chris@42 28 /* Generated by: ../../../genfft/gen_notw_c.native -fma -reorder-insns -schedule-for-pipeline -simd -compact -variables 4 -pipeline-latency 8 -n 4 -name n2fv_4 -with-ostride 2 -include n2f.h -store-multiple 2 */
Chris@42 29
Chris@42 30 /*
Chris@42 31 * This function contains 8 FP additions, 2 FP multiplications,
Chris@42 32 * (or, 6 additions, 0 multiplications, 2 fused multiply/add),
Chris@42 33 * 15 stack variables, 0 constants, and 10 memory accesses
Chris@42 34 */
Chris@42 35 #include "n2f.h"
Chris@42 36
Chris@42 37 static void n2fv_4(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@42 38 {
Chris@42 39 {
Chris@42 40 INT i;
Chris@42 41 const R *xi;
Chris@42 42 R *xo;
Chris@42 43 xi = ri;
Chris@42 44 xo = ro;
Chris@42 45 for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(8, is), MAKE_VOLATILE_STRIDE(8, os)) {
Chris@42 46 V T1, T2, T4, T5;
Chris@42 47 T1 = LD(&(xi[0]), ivs, &(xi[0]));
Chris@42 48 T2 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
Chris@42 49 T4 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
Chris@42 50 T5 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
Chris@42 51 {
Chris@42 52 V T3, T7, T6, T8;
Chris@42 53 T3 = VSUB(T1, T2);
Chris@42 54 T7 = VADD(T1, T2);
Chris@42 55 T6 = VSUB(T4, T5);
Chris@42 56 T8 = VADD(T4, T5);
Chris@42 57 {
Chris@42 58 V T9, Ta, Tb, Tc;
Chris@42 59 T9 = VSUB(T7, T8);
Chris@42 60 STM2(&(xo[4]), T9, ovs, &(xo[0]));
Chris@42 61 Ta = VADD(T7, T8);
Chris@42 62 STM2(&(xo[0]), Ta, ovs, &(xo[0]));
Chris@42 63 Tb = VFMAI(T6, T3);
Chris@42 64 STM2(&(xo[6]), Tb, ovs, &(xo[2]));
Chris@42 65 STN2(&(xo[4]), T9, Tb, ovs);
Chris@42 66 Tc = VFNMSI(T6, T3);
Chris@42 67 STM2(&(xo[2]), Tc, ovs, &(xo[2]));
Chris@42 68 STN2(&(xo[0]), Ta, Tc, ovs);
Chris@42 69 }
Chris@42 70 }
Chris@42 71 }
Chris@42 72 }
Chris@42 73 VLEAVE();
Chris@42 74 }
Chris@42 75
Chris@42 76 static const kdft_desc desc = { 4, XSIMD_STRING("n2fv_4"), {6, 0, 2, 0}, &GENUS, 0, 2, 0, 0 };
Chris@42 77
Chris@42 78 void XSIMD(codelet_n2fv_4) (planner *p) {
Chris@42 79 X(kdft_register) (p, n2fv_4, &desc);
Chris@42 80 }
Chris@42 81
Chris@42 82 #else /* HAVE_FMA */
Chris@42 83
Chris@42 84 /* Generated by: ../../../genfft/gen_notw_c.native -simd -compact -variables 4 -pipeline-latency 8 -n 4 -name n2fv_4 -with-ostride 2 -include n2f.h -store-multiple 2 */
Chris@42 85
Chris@42 86 /*
Chris@42 87 * This function contains 8 FP additions, 0 FP multiplications,
Chris@42 88 * (or, 8 additions, 0 multiplications, 0 fused multiply/add),
Chris@42 89 * 11 stack variables, 0 constants, and 10 memory accesses
Chris@42 90 */
Chris@42 91 #include "n2f.h"
Chris@42 92
Chris@42 93 static void n2fv_4(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@42 94 {
Chris@42 95 {
Chris@42 96 INT i;
Chris@42 97 const R *xi;
Chris@42 98 R *xo;
Chris@42 99 xi = ri;
Chris@42 100 xo = ro;
Chris@42 101 for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(8, is), MAKE_VOLATILE_STRIDE(8, os)) {
Chris@42 102 V T3, T7, T6, T8;
Chris@42 103 {
Chris@42 104 V T1, T2, T4, T5;
Chris@42 105 T1 = LD(&(xi[0]), ivs, &(xi[0]));
Chris@42 106 T2 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
Chris@42 107 T3 = VSUB(T1, T2);
Chris@42 108 T7 = VADD(T1, T2);
Chris@42 109 T4 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
Chris@42 110 T5 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
Chris@42 111 T6 = VBYI(VSUB(T4, T5));
Chris@42 112 T8 = VADD(T4, T5);
Chris@42 113 }
Chris@42 114 {
Chris@42 115 V T9, Ta, Tb, Tc;
Chris@42 116 T9 = VSUB(T3, T6);
Chris@42 117 STM2(&(xo[2]), T9, ovs, &(xo[2]));
Chris@42 118 Ta = VADD(T7, T8);
Chris@42 119 STM2(&(xo[0]), Ta, ovs, &(xo[0]));
Chris@42 120 STN2(&(xo[0]), Ta, T9, ovs);
Chris@42 121 Tb = VADD(T3, T6);
Chris@42 122 STM2(&(xo[6]), Tb, ovs, &(xo[2]));
Chris@42 123 Tc = VSUB(T7, T8);
Chris@42 124 STM2(&(xo[4]), Tc, ovs, &(xo[0]));
Chris@42 125 STN2(&(xo[4]), Tc, Tb, ovs);
Chris@42 126 }
Chris@42 127 }
Chris@42 128 }
Chris@42 129 VLEAVE();
Chris@42 130 }
Chris@42 131
Chris@42 132 static const kdft_desc desc = { 4, XSIMD_STRING("n2fv_4"), {8, 0, 0, 0}, &GENUS, 0, 2, 0, 0 };
Chris@42 133
Chris@42 134 void XSIMD(codelet_n2fv_4) (planner *p) {
Chris@42 135 X(kdft_register) (p, n2fv_4, &desc);
Chris@42 136 }
Chris@42 137
Chris@42 138 #endif /* HAVE_FMA */